Description of Study Fleet
By comparing manufacturer aircraft serial number data with FAA aircraft registration records, the NTSB identified 2,848 single-engine piston airplanes for the conventional cockpit display cohort and 5,516 for the glass cockpit cohort, all manufactured between 2002 and 2006.47 Figure 6 illustrates the rapidly changing distribution of the aircraft included in the study. Most aircraft in the conventional display cohort were manufactured between 2002 and 2004, while aircraft in the glass cockpit cohort first appeared on the FAA registry in 2003.
Figure 6. New registrations of aircraft study fleet by cockpit display configuration and year.
After 2004, the size of the conventional cohort remained relatively constant, while the size of the glass cockpit cohort increased rapidly, surpassing the conventional cohort in 2005 and nearly doubling it in 2006. Figure 7 shows the accumulated size of the study fleet for each year from 2002 through 2006 and the accumulated number of aircraft in the conventional and glass cockpit cohorts each year.
47 The study aircraft fleet was identified by comparing the aircraft serial number and cockpit display data
provided by manufacturers with FAA aircraft registry data. An aircraft was included in the study fleet if it ever appeared on the registry, regardless of whether it was subsequently deregistered or exported, or the registration later became inactive.
Figure 7. Accumulated study aircraft fleet size by cockpit configuration and year.
Description of Study Accidents
Study analyses of accidents included the years from 2002 through 2008. The years of aircraft manufacture from 2002–2006 were selected to correspond with the 2006 GAATAA 100-percent sampling methodology for newly built aircraft. However, although the activity and accident data included aircraft manufactured throughout 2006, the number of study aircraft did not stabilize until the end of that year. Accident records for 2007 and 2008 were therefore included to represent 2 full years of accidents associated with the study aircraft, unaffected by the addition of newly manufactured aircraft.
Accident Information
This section contains summary analyses of the relative accident occurrences and rates associated with the conventional and glass cockpit cohorts defined in this study. Comparisons of accident numbers are limited because aircraft can be used differently and in ways that expose one group of aircraft to more or less risk for severe accident outcomes than another. The validity of comparisons of accidents with aircraft manufacturing and registration records is likewise limited
by the possibility that aircraft may be sold, exported, deregistered, or placed in storage.48
Accident Involvement
Therefore, the accident data provided in this section should be considered within the context of corresponding activity data. Active aircraft information, flight activity, and aircraft usage data from the 2006 and 2007 GAATAA Surveys are presented throughout this section for comparison with FAA aircraft registry information and NTSB accident records from 2002 through 2008. Due to the unequal cohort sizes, the aircraft and accident data are presented as percentages of the respective cohort totals to facilitate interpretation. The complete list of study accidents is included in the appendix.
A comparison of the list of study aircraft with NTSB records identified 266 total accidents involving study aircraft between 2002 and 2008, 62 of which resulted in one or more fatal injuries.49
Comparisons of accident involvement from 2002 through 2006 must account for changes in the size of the study fleet due to the newly manufactured aircraft added to the fleet each year. The distribution of study aircraft and accidents associated with the glass cockpit cohort (as shown in figure 8) shows that the percentage of accidents involving glass cockpit aircraft was smaller than would be expected based on the percentage of the study fleet that those aircraft represented. During 2004 and 2005, fatal accidents for conventional and glass cockpit aircraft were proportional to the percentage of the study fleet they represented, but starting in 2006, the glass cockpit group began to experience proportionately more fatal accidents. Over the entire period from 2002 through 2008, aircraft in the glass cockpit cohort showed a disproportionately lower rate of total accidents per registered aircraft but a disproportionately higher rate of fatal accidents per registered aircraft than those in the conventional cohort.
Of the 266 study accidents, 141 accidents—23 of them fatal—involved conventionally equipped aircraft. The remaining 125 total accidents and 39 fatal accidents involved glass cockpit aircraft. It is important to note that direct comparisons of the overall accident totals would be misleading in this case because of the changes in the sizes of the cohorts during the time period analyzed.
48 For example, data provided by GAMA indicate that approximately 30 percent of aircraft produced in 2007
and 2008 were exported. Similar data do not exist for 2002–2006, but the number of aircraft that remained active in U.S. civil aviation is likely well below the 8,354 total aircraft that had appeared on the FAA aircraft registry. Further support for this suggestion comes from the active aircraft estimate results of the 2006 and 2007 FAA GAATAA Survey cited in table 2 of the Activity, Exposure Data, and Accident Rates section later in this chapter.
49 One accident (SEA03LA180) was excluded from study analyses because it occurred during the factory flight
a
The study aircraft fleet included aircraft manufactured between 2002 and 2006; therefore, the distribution of glass cockpit aircraft in the study aircraft fleet remained constant from 2006 through 2008.
Figure 8. Distribution of glass cockpit cohort aircraft and accidents per year.
Accident Severity
Statistical comparisons of the 2002 through 2008 accident data, independent of the registry or survey information, show similar differences in accident severity by cockpit display type. As shown in figure 9, the percentage of accidents resulting in fatality was about twice as high for the glass cockpit cohort as for the conventional cohort. Of the 266 accidents involving study aircraft between 2002 and 2008, accidents involving aircraft in the glass cockpit cohort were significantly more likely to be fatal: χ2
(1, N = 266) = 8.216, p = 0.004.50
50 Throughout this report, the results of chi-square (χ2) statistical tests are included in the text using the
following notation (degrees of freedom, N = number of cases compared) = resulting chi-square value, p = probability (that is, significance) of the result. The p value can be interpreted as the percent likelihood that the observed value occurred by chance. Therefore, a difference that results in a small p value is unlikely to have resulted from chance and is more likely the result of differences between the two groups. For the purposes of this study, p values of 0.05 (5 percent) and less are considered statistically significant. Refer to table 7 at the end of this chapter for a summary of all study chi-square statistical test results.
Figure 9. Comparison of study accidents by severity of outcome.
In addition to the statistical comparisons of accident data from 2002 through 2008, GAATAA Survey data were used to develop rates for the years 2006 and 2007 that represent accident risk by providing estimates of the number of aircraft actively operated, the number of hours flown, and specific characteristics of how those aircraft were operated. Rates based on flight activity provide a clearer indication of the relative safety of the conventional and glass cockpit configurations by identifying differences in the number of aircraft that were actively flying versus those that may have been sold or placed into storage, or that may have become inactive for other reasons. Differences between aircraft buyers who opted for a glass cockpit airplane can contribute to differences in accident risk, and pilots may use aircraft differently based on their avionics capabilities. The 2 years of available GAATAA Survey data were used to further explore these possibilities and provide evidence of activity and usage differences to aid interpretation of the aircraft’s accident record to date.51
Activity, Exposure Data, and Accident Rates
Of the 2,848 airplanes included in the conventional display cohort, 935 of the owners responded to the 2006 GAATAA Survey, and 472 responded to the 2007 survey. Of the 5,516 aircraft in the glass cockpit cohort, 1,803 owners responded to the 2006 GAATAA Survey, and 1,885 responded to the 2007 survey. These responses were used to calculate estimates of the number of active aircraft and total hours flown, as well as the number of hours flown by purpose of flight, day/night operations, and visual/instrument flight activity in accordance with the
51 Note that the study includes statistical comparisons of all accidents involving the study aircraft fleet from
2002 and 2008. However, GAATAA Survey data were only available for the years 2006 and 2007. No attempt was made to apply the 2006 and 2007 survey results to the prior years 2002 through 2005.
normal GAATAA Survey methodology.52
Table 1. GAATAA Survey analysis results.
Results of the active aircraft and flight hour analyses are presented in table 1 below.
Year Active Aircraft Flight Hours
Conventional Glass Cockpit Conventional Glass Cockpit
2006 2,412 4,203 593,853 805,152 (0.2% Std. Error) (0.2% Std. Error) (2.6% Std. Error) (1.6% Std. Error)
2007 1,738a 4,205 565,370 838,573 (0.2% Std. Error) (0.2% Std. Error) (4.0% Std. Error) (1.6% Std. Error) a
The large drop in the estimated number of active conventional aircraft in 2007 was likely due in part to increased variability in survey estimates for this group because aircraft manufactured during 2002 were not included in the 100-percent 2007 survey sample. This subsequently resulted in fewer total responses for aircraft built during 2002, which were exclusively of the conventional cockpit design. The effect of this change is that the number of conventionally equipped aircraft and their associated activity were likely higher than the survey results indicate for 2007. The effect of this change is also evident in the higher standard error value associated with the flight hours for the conventional cohort in 2007.
Dividing the flight hours from the GAATAA Survey by the number of active aircraft provides an estimate of the average number of hours flown per aircraft. For the years 2006 and 2007, the average estimated hours flown per aircraft was 286 for the conventional cohort (246 hours/aircraft in 2006 and 325 hours/aircraft in 2007) and 195 for the glass cockpit cohort (192 hours/aircraft in 2006 and 199 hours/aircraft in 2007).
As illustrated in table 2, accident rates calculated from NTSB accident records and GAATAA Survey data indicate that the 2006 and 2007 accident rates per 1,000 active aircraft were higher for the conventional display cohort, but the fatal rates were higher for the glass cockpit cohort for both years.
Because the study cohorts included only a few thousand aircraft, and the numbers of total and fatal accidents within the cohorts were relatively small each year, the 2006 and 2007 activity and accident data were summed for comparisons of accident rates and specific accident details to provide more stable rate estimates and to reduce the potentially distorting effect of small numbers of events on rate calculations. Even when using this approach, the standard errors associated with the fatal rates are high due to the relatively small number of total events.
52 Analyses of the subset of GAATAA Survey data were done by the contractor responsible for conducting the
survey in accordance with the established survey methodology; however, this is the first published use of a subset of GAATAA Survey data for targeted analyses like those included in this report.
Table 2. Accident rates for 2006 and 2007 per 1,000 active aircraft, by aircraft cockpit configuration.
Year Accident Rate Per 1,000 Active Aircraft
Total Fatal
Conventional Glass Cockpit Conventional Glass Cockpit
2006 9.12 7.85 1.24* 2.62
2007 12.08 6.9 1.15* 1.43*
Combined 2006-2007
10.36 7.37 1.20* 2.02
(15.3% Std. Error) (12.7% Std. Error) (44.9% Std. Error) (24.3% Std. Error) *Rate based on fewer than 10 events.
As previously stated, comparisons of accident rates based on active aircraft provide a better indication of risk than those based on numbers of aircraft manufactured because they do not include aircraft that were subsequently exported, placed in storage, or not flown. However, accident rates based on owner-reported flight activity provide the best indication of risk because they include both the extent of operation and the way the aircraft was operated. Accident rates calculated from the survey responses regarding the number of hours flown annually are shown in table 3, along with comparison accident rates for all general aviation operations.53
Those results indicate that the total accident rate per 100,000 flight hours was higher for the glass cockpit cohort in 2006, but higher for the conventional cohort in 2007. The combined 2-year accident rates per 100,000 flight hours for 2006 and 2007 were similar for both the glass and conventional cohorts (3.77 and 3.71 respectively). The total accident rate for both cohorts was less than the 6.63 accidents per 100,000 flight hours for all general aviation operations for the same period, reflecting the wide range of aircraft and flight operations included in general aviation.
The fatal accident rate for the glass cockpit cohort exceeded the rate of fatal accidents per 100,000 hours for the conventional cohort for both years, and for all general aviation operations in 2006. Like the rates per active aircraft discussed previously, the fatal rates per 100,000 flight hours for both cohorts—especially the conventional cohort—have large standard errors due to the small number of events. The resulting rates, however, are consistent with the results of other study analyses, indicating that accidents involving the glass cockpit cohort were more likely to be fatal. The combined fatal accident rate for 2006 and 2007 was higher for the glass cockpit cohort (1.03) than for the conventional cohort (0.43). The combined 2006 and 2007 fatal rates for both study cohorts were less than the 1.24 fatal accidents per 100,000 hours for all general aviation operations for the same period.
Table 3. 2006 and 2007 accident rates per 100,000 flight hours, by cockpit configuration.
Year Accident Rate Per 100,000 Flight Hours
Total Fatal
Conventional Glass Cockpit All General Aviation†
Conventional Glass Cockpit All General Aviation† 2006 3.70 4.10 6.33 0.51* 1.37 1.28 2007 3.71 3.46 6.92 0.35* 0.72* 1.20 Combined 2006-2007 3.71 3.77 6.63 0.43* 1.03 1.24
(15.3% Std. Error) (12.7% Std. Error) ― (44.2% Std. Error) (19.2% Std. Error) ― *Rate based on fewer than 10 events.
†
National Transportation Safety Board, Aviation Accident Statistics: Accidents, Fatalities, and Rates, 1989 - 2008, U.S. General Aviation. Available at: http://www.ntsb.gov/aviation/Table10.htm.
At the time of writing, targeted GAATAA Survey flight-hour data were not available to calculate 2008 accident rates for the study cohorts. If the hours flown for both cohorts during 2008 were similar to the averages in 2006 and 2007, the total accident rates for both the conventional and glass cockpit cohorts would again be less than the overall general aviation rate. However, with 13 fatal accidents during 2008, the fatal rate of aircraft in the glass cockpit cohort in this study would well exceed both the conventional cohort and the overall general aviation fatal rates for the year.54
Flight Conditions
Accident details and survey responses associated with both cohorts were compared to identify any differences in accident circumstances, aircraft use, or pilots that could affect the severity of accident outcomes and explain the observed differences between the accident and fatal accident rates for the conventional and glass cockpit cohorts. For example, accidents that occur at night or in instrument meteorological conditions (IMC) have historically been more likely to result in fatality than those that occur during the day in good weather.55
Time of Day
As illustrated in figure 10, the 2002 through 2008 accident data indicate that a higher percentage of accidents involving aircraft in the glass cockpit group occurred at night, but the difference was not statistically significant: χ2
(1, N = 266) = 3.058, p = 0.080.
54 For example, if total flight hours were estimated for the study groups by averaging the 2006 and 2007 survey
results, the 2008 fatal rate for the conventional group would be 0.35 fatal accidents per 100,000 flight hours, and the rate for the glass cockpit group would be 1.58, compared to the overall general aviation fatal rate of 1.20 in 2008.
55 Risk Factors Associated with Weather-Related General Aviation Accidents, Safety Study NTSB/SS-05/01
Figure 10. Comparison of study accidents by time of day.
As figure 11 shows, distribution of GAATAA Survey flight hour estimates by time of day for 2006 and 2007 was similar for both cohorts. However, when compared with the accident data for those years, the rates of total and fatal accidents per flight hour at night were higher for the glass cockpit cohort (see table 4).
Figure 11. Combined 2006 and 2007 flight hour distribution by time of day and cockpit configuration.
Table 4. Combined 2006 and 2007 accident rates per 100,000 flight hours by time of day and cockpit configuration.
Total Fatal
2006-2007 Conventional Glass Cockpit Conventional Glass Cockpit
Day 4.01 3.86 0.30 0.87
Night 1.85 3.31 1.23 1.84
Weather Conditions
As shown in figure 12, the 2002 through 2008 accident data indicate that a higher percentage of glass cockpit accidents occurred in IMC. The difference in accident weather conditions was marginally significant: χ2
(1, N = 264) = 3.639, p = 0.056.
Figure 12. Comparison of study accidents by weather.
GAATAA Survey data regarding flight hours by weather conditions, shown in figure 13, indicate that glass cockpit aircraft owners reported a larger percentage of flight time in IMC. A comparison of accidents and flight hours during 2006 and 2007 (table 5) shows similar total accident rates for both groups in visual meteorological conditions (VMC) but higher total and fatal accident rates per flight hour in IMC for the glass cockpit cohort.
Figure 13. Combined 2006 and 2007 flight hour distribution by weather and cockpit configuration.
Table 5. Combined 2006 and 2007 accident rates per 100,000 flight hours by weather and cockpit configuration.
Total Fatal
2006-2007 Conventional Glass Cockpit Conventional Glass Cockpit
IMC 1.63 2.68 1.63 2.34
VMC 3.86 3.94 0.29 0.67
Flight Plan Filed
Consistent with the previous results showing that glass cockpit aircraft spent a higher percentage of flight hours in IMC, the aircraft cohorts also differed with regard to flight plan filed for the accident flight. Figure 14 shows that among those accidents during 2002 through 2008 with flight plan information available, pilots in the glass cockpit cohort were significantly more likely to have filed an instrument flight rules (IFR) flight plan for the accident flight: χ2
(1, N = 250) = 11.718, p = 0.001. GAATAA Survey data do not provide estimates of the number of hours flown by type of flight plan.
Figure 14. Comparison of study accidents by flight plan filed.
Purpose of Flight
The study cohorts differed noticeably with regard to aircraft usage. Figure 15 shows that accident flights involving aircraft in the conventional cohort were almost equally split between instructional flights and personal/business flights, while glass cockpit accidents were significantly more likely to involve personal/business flights: χ2
Figure 15. Comparison of study accidents by purpose of flight.
GAATAA Survey results regarding reported aircraft use indicated that a larger percentage of the conventional cohort’s activity during 2006 and 2007 involved instructional flights, while the glass cockpit aircraft were more often used for personal and business flying (see figure 16). A comparison of accidents with reported aircraft use from 2006 through 2007, summarized in table 6, indicates that the conventional aircraft experienced higher total accident rates during both instructional and personal/business flying. Both cohorts experienced equally low fatal accident rates for instructional flights, but the glass cockpit cohort experienced a higher fatal accident rate during personal/business flights.
Figure 16. Combined 2006 and 2007 flight hour distribution by purpose of flight and cockpit configuration.
Table 6. Combined 2006 and 2007 accident rates per 100,000 flight hours by purpose of flight and cockpit configuration.
Total Fatal
2006-2007 Conventional Glass Cockpit Conventional Glass Cockpit
Instructional 3.98 2.79 0.20 0.20
Personal/Business 6.62 5.05 1.26 1.65
Planned Length of Flight
Among those accidents for which both point of departure and intended destination were known, the median planned length of accident flights associated with the glass cockpit cohort was 96 nautical miles (nm), compared to a median of 25 nm for conventional aircraft flights. Differences in the planned length of study flights for both cohorts were evaluated using the Mann-Whitney U test statistic. Results indicated that accident flights involving the glass cockpit cohort were significantly longer than those for aircraft in the conventional cockpit cohort (U = 5649.5, N (conventional) = 140, N (glass cockpit) = 122, p < 0.001).56
56 Throughout this report, the results of Mann-Whitney U statistical tests are included in the text using the
following notation (calculated value of the U statistic, n1 and n2 = number of cases in each study cohort, p =